[c-nsp] protecting cisco switches

[Mark Messier]

A site I manage has top-of-rack HP switches redundantly connected to
two cisco 4948 switches, and those 4948 switches are interconnected
with a trunk.  Every now and then an HP switch appears to do something
stupid and this impacts the cisco switches, sometimes fatally.  This
can happen with no one stirring the pot.

The last event was a bunch of %C4K_EBM-4-HOSTFLAPPING which crushed
the c4948 in short order.  This happened with no concurrent human
activity... the offending loop (yes, there was a loop, different than
the desired redundancy loop described above) was "installed" about ten
days prior.

What can I do on the cisco switches to protect them
and yet still leave them functional enough to achieve
the desired redundancy?   (and I'm not looking for social
engineering solutions)

Thanks,
-mark